AI for Dermatology | Opportunities, Risks, Challenges and the Future

Listen or watch on Spotify here.

We spoke with Dr Dilraj Kalsi, Clinical AI Lead at Skin Analytics, and Dr Alice Plant, Consultant Dermatologist at University Hospitals Dorset NHS Foundation Trust, about the mounting pressures faced by UK dermatology departments and how AI-powered technology could provide a solution. Our discussion specifically focuses on Skin Analytics’ DERM—the first and only autonomous AI in the cancer space.

Dermatologist Supply Falls Short of Growing Skin Cancer Cases

Skin cancer rates have grown almost three-fold since the 1990s, with an estimated 17,500 new melanoma cases diagnosed in the UK every year. This marks it as the fastest-growing common cancer type.

While survival rates are generally good —especially if caught early— skin cancer mortality is rising in parallel with incidence. According to Cancer Research UK, the number of deaths has increased by 141% over the last three decades, compared with a 147% increase in cases. Although the reasons behind this are multifaceted —including more international travel to sunnier locations and the aging population— dermatology shortages in the UK are likely contributing to the growing death toll.

One study recognizes the “tsunami of skin cancers” as having drastically increased dermatologist workloads, taking up 50% of clinicians’ time. Still, this was in 2014—a decade before melanoma cases reportedly reached an all-time high in 2024.

Now, with urgent referrals for skin cancer growing by 10% every year, UK dermatology departments are in crisis. About one in four dermatology posts remain unfilled, dwindling the NHS workforce and causing waitlists to creep up.

And the effect is palpable. In 2024, over 100,000 patients had to wait >28 days for a diagnosis after GP referral, with 39% waiting for more than 18 weeks.

Not only does this delay treatment for skin cancer patients —where intervening at an early stage can mean the difference between life or death— but it also worsens outcomes for those with non-cancerous inflammatory conditions.

“We just haven’t got enough doctors to see all those patients,” says Dilraj. “When you get these longer and longer wait times, there are a lot of patients who will experience deteriorating quality of life.”

Training More Dermatologists is Not the Answer

To tackle the issue, it might seem intuitive to simply train more dermatologists from the UK’s pool of practicing physicians—for example, teaching GPs to diagnose and treat skin cancers in-house. However, Dilraj outlines that this is not a feasible fix, noting that “GPs are already overstretched as it is” and that “this solution alone is not enough to make up the shortfall.”

As of 2019, there were 818 consultant dermatologists in the whole of the UK—roughly equal to one for every 79,200 people. With the President of the British Association of Dermatologists (Dr Tanya Bleiker) stating that a ratio of one for every 55,000 would be “ideal”, an extra 400 dermatologists is needed—almost half of the UK’s current capacity.

And, although the need is pressing, enrollment in specialist dermatology training is not reflecting current demands, as NHS priorities are pulled in other directions.

“As a whole, we need more people in acute hospitals, and we need more GPs,” explains Alice, building on Dilraj’s point. “It’s not anytime soon that we’re going to get more of those training dermatologists.”

She adds:

“The growing rate of skin cancer, plus fewer trained primary care staff and not enough staff within secondary care—it’s a recipe for disaster.”

Skin Analytics: Safely Triaging Patients Away from Urgent Referral Paths

At the forefront of the problem, Skin Analytics are supporting NHS dermatology departments with their AI-powered solution: Deep Ensemble for the Recognition of Malignancy (DERM).

With skin lesions, GPs tend to err on the side of caution and refer their patient to a dermatologist, who will perform further tests to confirm a diagnosis.

While caution is important for catching as many cancer cases as possible, urgent referral pathways are overwhelmed with patients who don’t really need to be there. One 2025 study, published in Cureus, found that only 13% of cases flagged as “suspicious” by GPs turned out to be skin cancer.

DERM significantly reduces these unnecessary referrals by quickly identifying benign cases as suitable for discharge, freeing up to 95% of face-to-face NHS urgent skin cancer appointments.

To use it, healthcare professionals just need to take close-up photos of the lesion using a smartphone fitted with a dermoscopic attachment. The AI then classifies the lesion as cancerous, precancerous, or something harmless, assigning a level of urgency to a patient in the process.

Benign cases receive an all-clear letter confirming the results, whereas patients with suspected skin cancer are connected virtually to a dermatologist who will map out next steps—again, avoiding  face-to-face appointments unless absolutely necessary.

The new pathway lends itself to a new wave of digital dermatology care, often referred to as “teledermatology.” Alice points out that, although experts have “toyed with the concept for quite some time,” teledermatology has come a long way in a short period. Just a few years ago, nurses would take photos of skin lesions to later send to dermatologists, trying to cut down time spent moving patient-to-patient. However, these images were captured with non-specialist equipment, leaving doctors to scrutinize poor-quality photos without the support of AI.

Now, DERM fills in those gaps, polishing the field of teledermatology with AI so that the crisis can be tackled head-on:

“I now have a virtual clinic where I can get through at least double the number of patients I could if they were face-to-face appointments,” says Alice. “I can easily review the images and the history of the skin growth, and then either discharge the GP with management advice or list the patient directly for surgery.”